From 53f57e75c6ebd4065643f951537eb8816cd2d0c3 Mon Sep 17 00:00:00 2001
From: Jean-Francois Dockes <jfd@recoll.org>
Date: Sat, 21 Nov 2015 12:48:09 +0100
Subject: [PATCH] rcldb syngroups: allow setting synonyms source after
 construction

---
 src/rcldb/rcldb.cpp    |  8 +++++---
 src/rcldb/rcldb.h      | 14 ++++++++++++--
 src/rcldb/rclterms.cpp |  8 +-------
 3 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/src/rcldb/rcldb.cpp b/src/rcldb/rcldb.cpp
index 87da3870..c9c412c9 100644
--- a/src/rcldb/rcldb.cpp
+++ b/src/rcldb/rcldb.cpp
@@ -764,9 +764,6 @@ bool Db::open(OpenMode mode, OpenError *error)
     }
     if (!m_config->getStopfile().empty())
 	m_stops.setFile(m_config->getStopfile());
-    // Synonyms are only used at query time for now
-    if (mode == DbRO)
-	m_syngroups.setfile(m_config->getSynGroupsFile());
 
     string dir = m_config->getDbDir();
     string ermsg;
@@ -1241,6 +1238,11 @@ void Db::setAbstractParams(int idxtrunc, int syntlen, int syntctxlen)
 	m_synthAbsWordCtxLen = syntctxlen;
 }
 
+bool Db::setSynGroupsFile(const string& fn)
+{
+    return m_syngroups.setfile(fn);
+}
+    
 static const string cstr_nc("\n\r\x0c\\");
 #define RECORD_APPEND(R, NM, VAL) {R += NM + "=" + VAL + "\n";}
 
diff --git a/src/rcldb/rcldb.h b/src/rcldb/rcldb.h
index 2d679648..38f34e39 100644
--- a/src/rcldb/rcldb.h
+++ b/src/rcldb/rcldb.h
@@ -456,6 +456,9 @@ class Db {
     }
     bool doFlush();
 
+    // Use empty fn for no synonyms
+    bool setSynGroupsFile(const std::string& fn);
+    
     /* This has to be public for access by embedded Query::Native */
     Native *m_ndb; 
 private:
@@ -477,13 +480,20 @@ private:
     // First fs occup check ?
     int         m_occFirstCheck;
 
+    // Synonym groups. There is no strict reason that this has to be
+    // an Rcl::Db member, as it is only used when building each It
+    // could be a SearchData member, or even a parameter to
+    // Query::setQuery(). Otoh, building the syngroups structure from
+    // a file may be expensive and it's unlikely to change with every
+    // query, so it makes sense to cache it, and Rcl::Db is not a bad
+    // place for this.
+    SynGroups m_syngroups;
+
     /***************
      * Parameters cached out of the configuration files. Logically const 
      * after init */
     // Stop terms: those don't get indexed.
     StopList m_stops;
-    // Synonym groups
-    SynGroups m_syngroups;
 
     // Truncation length for stored meta fields
     int         m_idxMetaStoredLen;
diff --git a/src/rcldb/rclterms.cpp b/src/rcldb/rclterms.cpp
index bd2eebba..5631fde2 100644
--- a/src/rcldb/rclterms.cpp
+++ b/src/rcldb/rclterms.cpp
@@ -270,16 +270,10 @@ bool Db::termMatch(int typ_sens, const string &lang, const string &_term,
                 exp1.swap(lexp);
                 sort(lexp.begin(), lexp.end());
                 lexp.erase(unique(lexp.begin(), lexp.end()), lexp.end());
-                LOGDEB(("ExpTerm: stemexp: %s\n", 
+                LOGDEB(("ExpTerm: stemexp: %s\n",
                         stringsToString(lexp).c_str()));
             }
 
-            // Expand the result for synonyms. Note that doing it here
-            // means that multi-term synonyms will not work
-            // (e.g. stakhanovist -> "hard at work". We would have to
-            // separate the multi-word expansions for our caller to
-            // add them as phrases to the query. Not impossible, but
-            // let's keep it at single words for now.
             if (m_syngroups.ok() && (typ_sens & ET_SYNEXP)) {
                 LOGDEB(("ExpTerm: got syngroups\n"));
                 vector<string> exp1(lexp);